linux-kernel - Re: Recent changes mean sb_min

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Z_VZT9UhPdLT-rk_@bombadil.infradead.org>
Date: Tue, 8 Apr 2025 10:13:51 -0700
From: Luis Chamberlain <mcgrof@...nel.org>
To: Jan Kara <jack@...e.cz>
Cc: Phillip Lougher <phillip@...ashfs.org.uk>,
	Andreas Gruenbacher <agruenba@...hat.com>,
	Namjae Jeon <linkinjeon@...nel.org>,
	Sungjong Seo <sj1557.seo@...sung.com>,
	OGAWA Hirofumi <hirofumi@...l.parknet.co.jp>,
	Carlos Maiolino <cem@...nel.org>,
	"Darrick J. Wong" <djwong@...nel.org>,
	linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: Recent changes mean sb_min_blocksize() can now fail

On Tue, Apr 08, 2025 at 12:51:37PM +0200, Jan Kara wrote:
> On Tue 08-04-25 12:39:52, Jan Kara wrote:
> > Hi!
> > 
> > On Tue 08-04-25 06:33:53, Phillip Lougher wrote:
> > > A recent (post 6.14) change to the kernel means sb_min_blocksize() can now fail,
> > > and any filesystem which doesn't check the result may behave unexpectedly as a
> > > result.  This change has recently affected Squashfs, and checking the kernel code,
> > > a number of other filesystems including isofs, gfs2, exfat, fat and xfs do not
> > > check the result.  This is a courtesy email to warn others of this change.
> > > 
> > > The following emails give the relevant details.
> > > 
> > > https://lore.kernel.org/all/2a13ea1c-08df-4807-83d4-241831b7a2ec@squashfs.org.uk/
> > > https://lore.kernel.org/all/129d4f39-6922-44e9-8b1c-6455ee564dda@squashfs.org.uk/
> > 
> > Indeed. Thanks for the heads up!
> 
> But isofs is actually fine since setting bdev block size needs exclusive open
> (i.e., has to happen before filesystem mount begins and claims bdev) and
> isofs does:
> 
> 	if (bdev_logical_block_size(s->s_bdev) > 2048)
> 		bail
> 
> in its isofs_fill_super().

Regardless, we added commit a64e5a596067bddb ("bdev: add back PAGE_SIZE
block size validation for sb_set_blocksize()" to effectively revert back
to the original behaviour, and so only filesystems which have FS_LBS are
not blocked by PAGE_SIZE.

Let me know if you still are seeing issues even with this patch merged.

  Luis